Skip to main content

Imagination On The Move 2013: PowerVR, MIPS, And More

MIPS: Now Brought To You By Imagination

The past few years have given me no shortage of opportunities to draw parallels between MIPS and Kodak. You see, both companies helped shape their respective industries, yet were ultimately surpassed by their competition and exiled to the annals of history. Consequently, Imagination's decision to acquire MIPS for $100 million in December 2012 and subsequently license nearly 500 of the company's patents did come as a considerable surprise.

Speaking about the due diligence process before the acquisition, CEO Sir Hossein Yassaie explained that two main factors fueled the ultimate decision to go through with it.

First, he stated that MIPS processors offer the cleanest "true RISC" architecture and that they had "maintained elegance" for over 20 years, facilitating superior performance in a smaller size with unmatched low-power credentials.

Second is the belief that ARM's entrenchment presents a danger of stifling the CPU landscape (perhaps similar to what we're seeing in the desktop space), and that the market requires "balance and credible choices" to reach its full potential. Hence, his oft-repeated adage that:

"The industry needs MIPS as much as MIPS needs the industry, because no industry can operate in an environment where it is a near-monopoly. I often say on this subject that wherever there is a Coke, there will be Pepsi. And MIPS is here to make sure that there is a Pepsi."Source: Engadget

With regards to Imagination's objectives for MIPS, the company aims to ensure that people feel confident about the future of MIPS, and ultimately develop a family of class-leading processors that combine the underlying values of MIPS' architecture with Imagination's deep understanding of ultra-low-power and heterogeneous processing. These efforts be seen first in the MIPS Series 5 "Warrior" processors, though the company eventually aims to target any device requiring a CPU, from smartphones to servers and wearable tech.

MIPS Series 5 "Warrior" Processors

Although Imagination has yet to provide full technical specifications for those next-generation processors, the company confirmed that it plans to offer binary 32- and 64-bit compatibility across the full range, in addition to promising the best performance and lowest power consumption in a given die area. The family will also incorporate the MIPS SIMD Architecture (MSA), which it says will help developers leverage old code as they write new software, provide a highly-scalable security framework for content protection and networking protocols, and hardware multi-threading in select cores.

The MIPS Series 5 "Warrior" cores should arrive this year, with availability for all product devices and markets tentatively scheduled for the end of 2014.

  • dgingeri
    "There should always be Pepsi wherever there is Coke," it was said, and Imagination wants to be Pepsi.

    With AMD failing so spectacularly in the CPU arena. It's nice to see someone else stepping up to the plate.
  • ronch79
    Should be interesting how Imagination's MIPS strategy pans out. I'm getting tired of the x86/ARM virtual duopoly.
  • Textfield
    I'm really interested to see how their GPUs pan out. They're one of the few companies that's actually attempting commercially available ray-tracing hardware. It hasn't had much of a use outside the professional space for a while, but game engines are just starting to make use of ray-tracing to calculate more advanced lighting effects. This little hardware tweak could bring mobile graphics to the next level.

    I also like the idea of having the compute sector separate from the graphics sector. As it is now, the unified shader architectures of modern GPUs require that devs choose between graphics and compute. That choice is made even more complicated when the OS has to handle the conflicting requests for graphics and compute from all the apps currently running.
  • Matthew Carlino
    "There should always be Pepsi wherever there is Coke," By the looks of things I guess that makes AMD RC cola?
  • somebodyspecial
    They will be out of business or bought in 5yrs not owning 25% of any market. They had to borrow 20mil just to make the 100mil purchase of mips which BTW was their only option I guess. They didn't have the in-house staff to do a cpu at this stage (even a new team would take what like 2-3yrs to put a soc out with their own cpu?) so had to buy mips. Meanwhile they are fighting ARM themselves (~750mil), Qcom (40bil), NV (3.5B), Intel (10.5B, but just raised ~5Bil in bonds ~dec last year, first time in a LONG time, a sign of weakness at intel keeping up with other fabs IMHO, they can only fund their dividends for another 2-3yrs etc with current fab costs etc), samsung (Billions, making 7.5B/Q now dwarfing even Intel, thus about to turn tides in fabs) and maybe you can consider apple also (since they now own UMC and already make their own cpus and will likely use them in macs soon, maybe a PC chip coming too?).

    When looking up companies and cash I was surprised to find out ARM doesn't even have 1B in cash to fight this war. Clearly IP licensing doesn't make you rich (powervr makes less than 20mil on their gpus being in every single apple phone/tablet etc...ROFL), and I think this is an NV/QCOM/INTEL/Samsung race with all others losing. I'd also say NV wins a lot soon as the gpu is becoming KING over cpus. Only AMD and NV have the gaming chops to win that war with 20yrs of working with devs and 20yrs of getting drivers out the door for every game that comes along (obviously AMD being FAR weaker, with 3B debt, out of cash this year or Q1 etc if they don't turn the corner).

    Everyone else has to play catch up in games with NV/AMD. IF AMD can make it until 2015 they have a chance at getting back in the game (when they get a SOC out). Consoles won't save AMD with single digit margins for a while and no guarantee they will sell past xmas very well (100mil tops in next 12months off that, won't even pay GlobalFoundry bill coming Dec and those profits will take 12months to get IF consoles sell 10mil/each for xbox1 and ps4 during that time), and their cpus (steamroller etc) still look like they'll trail Intel by large margins in perf thus again selling dirt cheap vs. Intel. GPU's will be held down by NV (they can price cut you to death if needed, as they have no debt and can give away gpus at break even for years with 3.5B in the banks). Only SOCS can really save AMD in the next few years, as I don't think consoles have a chance once 16/14nm hits they will be caught in a tablet most likely or good enough to make them pointless and massive game quantities will be on android by then (hawken types, not angry birds, 60% of devs are making mobile games). 14nm/16nm should be here in 2016 (early most likely) and consoles will be caught by shield r3-r4 and all it's competition (ouya rev3/4, google rev2-3, apple rev2/3, amazon rev2-3 etc. all coming with xmas consoles or early 2014, so all will be rev2 or 3 2yrs later). The largest number of consoles are sold in year 3-8 when casual gamers join once there is a huge library of games out. They won't buy this time as they'll all have GOOD ENOUGH phones/tablets or SOC consoles by then.

    I hope someone buys AMD that owns fabs (samsung, apple? TSMC?) but I'm not sure they would be allowed to produce x86 if that happens. Can Intel claim ARM is entering their space in desktops/notebooks/servers already or next year and kill x86 in this case, or can they just do it anyway legally? Not sure about that but it seems pretty clear AMD & IMG.L are weaklings in this game. I also see QCOM getting hurt soon (already margins are sinking outside USA) due to their modems not being the THING to have anymore, and now turning to GPUS while everyone has a modem now or coming next year (again IMG.L and AMD have no modem, which is why TI exited, NV got Icera first, TI missed the boat).

    Broadcom probably gets hit here too, lacking gpus and anything special now (qcom boat, modems are over now, time for a cpu/gpu soc world). They bought netlogic for 3.7B 2012, but have a shrinking cash pile last I checked and Cisco fighting for network crap, while qcom etc eat at modems. Not sure how they grow from here but maybe they could buy AMD? Surely should be cheap soon if not already...LOL. I don't think they have more than ~1B left (after taking out debt) but could easily borrow to complete a purchase of AMD for say 2-4B etc (or merge?). Not a bad play considering they'd have a cpu, gpu, ethernet, networking crap, modems, etc then. Well rounding company at that point but again can they make x86 in that case?

    Just some stock observations gathering data. I would not invest in IMG.L :) No rogue 6 benchmarks either, and I don't think compute will mean anything in a phone/tablet for a while which is just wasted soc space to me. NV went right by cutting features to get great watts/perf with T4/T4i/T5. Features that you can't implement without bogging down fps are pointless (like DX11.1 on a soc at 28nm...LOL, most desktops can't handle it at 1920x1200 etc with 100watt+ cards).

    Once T4i hits Qcom margins will take further hits outside USA (maybe inside too?) and they still don't have a gaming site like tegrazone or any games to brag about yet, so I'm thinking this eventually hurts INSIDE USA too. There may still be some upside but I see another apple situation happening (no place but down until something changes in games IMHO).

    Xiaomi Phone 3 in china using T4 and Snapdragon and more of this will happen as T4i hits. This phone just hit sept. They've only been in business for 3yrs and already bagged 15mil+ phones and expecting 20mil next year. No doubt they'll be selling T4i in china soon (xiaomi is 6th in china phone sales).

    Any thoughts people? :)
  • somebodyspecial
    T4 phone article...Welcome to 745mil subscriber market NV. I wonder how many they'll get? :)

    My bad, forgot with share buybacks and dividend NV dropped to 3B not 3.5B in cash. STill a lot vs. IMGL, Broadcom, AMD.
  • derekullo
    Uncle Brody could sell his MacBook Air and buy another Chevrolet Corvette Stingray Convertible ...
  • Travis Hershberger
    As someone who cut his teeth on SGI MIPS hardware having more MIPS based devices around would be great. I would love to see the platform advance and compete in today's marketplace.
  • ftw2
    It's an interesting industry that's for sure. Having seen many companies come and go over the years it's hard to really know what will happen.

    I think you're right that the small players are in big trouble. I'm not as optimistic about nvidia's short term chances as I think the big brands aren't currently using them (apple, samsung, amazon, google (other than 1st gen nexus 7))

    As for AMD... I don't know what their future holds. Their GPU tech is good, drivers are pretty good, etc, but their cpu's are in a dire situation. It's very hard to find a scenerio that an AMD CPU is the ideal choice if you factor price out. They're cheap and 'good enough' for many things but hardly anything to get excited about. I've owned a lot of AMD and ATI/AMD products over the years, still own many AMD GPU's, but it's a company that's evolving too slowly I think.

    So my predictions are. Intel maintains windows stranglehold. Apple and Samsung design all their own gpu/cpu/etc. Amazon/Google continue to use qualcomm/intel/nvidia whoever happens to be the best at the time they need it. Microsoft/Amazon/Google may buy one of those hardware/chip companies. MS clearly feels the only way to compete with apple and samsung is to do it all yourself, after the nokia buyout, I think an nvidia buyout may make sense. They make gpus that work with windows and have a mobile portfolio.
  • Murissokah
    11494566 said:
    "There should always be Pepsi wherever there is Coke," it was said, and Imagination wants to be Pepsi.

    With AMD failing so spectacularly in the CPU arena. It's nice to see someone else stepping up to the plate.

    Dude, there's just so much wrong with that statement. Imagination can't be "stepping up to the plate" because they are on an entirely different market... and also because PowerVR has been around for quite a while.

    AMD is not "failing spetacularly" in the CPU front, unless you value benchmarks more than corporate strategy. They certainly can't compete for top performance and fail to deliver what we came to expect from them in the past, but that hardly means they failed. They focus on different things, as we can see from them managing to outsmart the competition in the console front. I'm an Intel guy ever since P2, but this is just plain bashing.

    As for the real topic, which has nothing to do with AMD, Imagination has an uphill battle against the plethora of ARM SOCs. Then again, this is a market so big that a far second place still makes you a billionaire.